Two of the main areas involving these techniques are space-time processing (smart antennas) and multi-user detection (MUD). An ST-MUD receiver architecture is presented and the performance of the architecture with a minimum mean square error (MMSE) decision criterion is analyzed in a frequency selective Rayleigh fading channel.
Introduction
Problem formulation
ST-MUD for MC-DS-CDMA includes all these signal processing related techniques to enable 3G communication systems. The performance of the ST-MUD architecture, when using subspace techniques, is therefore also important.
Outline of thesis
A summary of the thesis and concluding remarks are made in Chapter 7, along with some ideas for future research. Appendix C summarizes the details relevant to the implementation and configuration of the custom simulation environment, and its results.
Original Contribution
In Appendix B, the elements of the multiaccess-interference-plus-noise covariance matrix of Chapter 5 are derived.
Vector Channel Models and Propagation
- Introduction
- Mobile Radio Propagation
- Propagation Loss
- Shadowing (Log-normal slow fading)
- Fading
- Frequency Selective Fading
- Frequency Diversity Model
- Vector/Spatial Channel models
- Geometric Models
- Statistical Vector Channel Models
- Measurement based
- Fading Correlation
- Space model
- Joint Space-Frequency Model
- SUlTIlTIary
Measurements have shown that the average value of the received signal level has a lognormally distributed value. The Doppler power spectrum (R(A)) of the signal quantifies the time-varying properties, such as the fading rate, of the received signal.
Smart Antennas
- Diversity
- Maximal Ratio Combining
- Switched bealTI
- Beamforming
- Beamforming array geometries
- Array Performance Criteria
- Optimal Processing and the Wiener Solution
- Beamforming Algorithms
- Combined Space-time Processing
- Summary
With zero routing and beamforming, it is possible to increase the signal-to-interference ratio of the desired user. These smart antennas use algorithms that take into account the temporal structure of the signal. The ahlal radiation pattern of the antenna array is unchanged, but by carefully combining the signals on the different antenna elements, the SlNR of the desired user signal can be increased.
An equal gain combiner combines the signals from the desired user at the various antenna elements with an equal weight. A selection diversity scheme will choose to receive the signal from only one of the antenna elements according to a certain criterion. Maximum ratio combining combines the signals from the desired user with a weighting factor proportional to the SNR at each antenna element.
The signal processing unit continuously adjusts the radiation pattern of the antenna array to maximize a specific performance measure (eg SINR of the desired user). Adjusting the phase shift (or complex weights) on each antenna element (as in Figure (3-1)) changes the radiation or beam pattern of the antenna array. The phase shift between two antenna elements is a function of AOA and the distance between sensors d, and is.
Multiuser Detection
DS-CDMA Model
An asynchronous DS-CDMA transmitter model for the uplink of a mobile radio network is considered. The baseband representation of the kth user's transmitted signal is given by where Ak and Sk indicate the amplitude and the normalized dispersion waveform for the kth user respectively, T is the duration of the data symbol, r'k is the start time of the user k. The transmitted symbol bdi) of the kth user assumes the values {+I,-I} with equal probability. The output from a bank of matched filters, each with the respective knowledge of the correct timing, phase and spreading sequence for all users is a sufficient statistic to perform optimal MUD, and as such it is convenient to write the vector r of these values in the shape.
The noise vector n consists of independent and identically distributed (i.i.d.) zero-mean Gaussian random variables with covariance matrix 0'21, where 0'2 is the variance of the noise with a power spectral density equal to t no.
Performance Measures
The multi-user efficiency ratio is then defined as the ratio between effective and actual energy. When the MUD works very well, the effective energy is very close to the actual transmitted energy, so the multi-user efficiency ratio approaches one. When the MUD performs poorly, it means that the transmitter has to transmit at a very high level to achieve the required BER.
The multi-user efficiency ratio quantifies the performance loss due to the presence of other users, and therefore the reciprocal (in dB's) of the multi-user efficiency ratio, is a positive value and is often referred to as the degradation factor. Near-far resistance is defined in [10] as the asymptotic multiuser efficiency minimized over the received energies of all the other users, and is . The near-far resistance is useful for calculating how effectively a receiver can overcome the near-far problem.
OS-COMA systems spreading rate (or processing gain) is directly proportional to bandwidth, thus another interpretation of spectral efficiency would be the number of bits per chip that can be reliably transmitted (biIS.chip- I). The increased spectral efficiency of the channel can thus be used to evaluate the relative performance of MUD techniques.
Types of Multiuser Detection
- Linear Multiuser Detection
- Interference Cancellation
They fall under the category of one-shot detectors, although some linear techniques, such as MMSE, can be implemented adaptively. This means that its performance is completely independent of the power conditions of the users in the system, i.e. The MMSE solution takes into account the SNR of the different users in the system, and produces a solution that is superior to that of the decorrelating detector.
Another important advantage of the MMSE technique is the ease with which it can be implemented adaptively [9). The near-far resistance and the asymptotic multiuser efficiency of the MMSE solution are therefore the same as for the decorrelating detector (4.25). The leakage coefficient fJk is the contribution of the kth user to the decision statistics for user I.
After each iteration, no hard decisions are made about the bits transmitted, but rather cautious, soft decisions. The multi-user asymptotic efficiency proved to be a useful metric for comparing the performance of different MUD schemes. An abbreviated performance analysis of the MMSE technique was conducted as it provides insight into the expected behavior of the MMSE ST-MUD detector.
Space-Time MUD
- Introduction
- MC-OS-COMA Model
- Channel Model
- MMSE ST-MUD Receiver ModeL
- Received Vector of Samples
- MMSE Solution
- Performance Analysis
- Results
- Reduction of Multiaccess Interference
- Sumnlary
The received vector of samples is fundamental to the formulation of the joint space-frequency multipath MMSE solution that the ST-MUD architecture uses. The amplitude, A, of the desired signal component at the output of the receiver is therefore. Antenna alTay improves the performance of ST -MUD when there is uncollected fading by providing diversity.
A common receiver architecture consisting of 3 antenna elements, 3 subcarriers and 3 multipath receivers is used to analyze the ST -MUD performance. Figure 5-8 shows that the ST -MUD simulation and analysis agree, which validates the analysis technique. The initial performance gap is due to the increased diversity caused by the ST-MUD antenna array and multipath receivers.
Increasing the number of multipath receivers has the most dramatic effect on system capacity. MAl suppression capability of ST -MUD was illustrated in Figures 5-4, 5-5 and the increased diversity in 5-6. Figures 5-8 and 5-9 showed the performance gain of the ST-MUD architecture over the conventional matched-filter receiver.
Subspace Techniques
- Introduction
- Subspace Projection
- Principle Components
- Cross-Spectral Filterin g
- Partial Despreading
- Results
- Sumnlary
The partial despreading technique provides a mechanism to scale the performance of the receiver from that of the full-rank MMSE solution to that of the conventional matched filter. The conventional matched filter and MMSE receiver performance are shown as upper and lower limits for the performance of the receivers. The cross-spectral technique converges faster than the principal component technique, and as the rank of the receiver approaches the number of users, the cross-spectral technique converges to the performance of the MMSE solution.
The performance of the partial despreading technique falls between the matched filter and the MMSE solution, depending on the despread factor. It is readily apparent that the subspace techniques allow a reduction in the number of filter coefficients while still achieving a performance comparable to that of the full MMSE receiver. The relative performance of the subspace techniques when they are all of the same rank is further illustrated in Figures 6-5 and 6-6.
In the higher SNR region, the performance of the subspace techniques deviates from the MMSE performance due to their reduced MAl suppression capability. The performance of subspace teclmiques was found to be highly dependent on the choice of dimension (or rank). The partial despreading technique provided a mechanism to scale receiver performance and complexity between the matched filter receiver and that of the full MMSE receiver.
Conclusion
Dissertation Summary
MAl suppression, near-far resistance, and overall receiver structure of the MMSE MUD scheme for OS-COMA communication systems were illustrated. The binding detector was covered since some of the performance metrics of the MMSE scheme derive from it. The MMSE scheme was important as it provided the framework to derive the MMSE ST-MUO detector of the next chapter.
The implementation of decentralized MMSE MUD using an FlR digital filter along with an adaptive algorithm was presented, which provided a mechanism for realizing MMSE ST-MUO. Using the receiver structure and the vector impulse response of the chalU1el, the received vector of samples was obtained. The joint space-frequency-multipath MMSE solution was derived, which allowed the performance analysis of MMSE ST-MUO.
The simulation and analysis results are in agreement and illustrate the suppression capability of the MAl receiver. The use of subspace techniques was motivated and their weaknesses and trade-offs were covered. We investigated the effect of rank selection, number of users and SNR.
Future Directions
Pettersen, “A Review of Smart Antenna Technology for Mobile Communications Systems,” IEEE Communications Surveys, pp. Haimovich, “Performance Analysis of Optimal Combination in Wireless Communications with Rayleigh Fading and Cochannel Interference,” IEEE Trans. Calderbank, “Space-time codes for high-data wireless communications: performance criterion and code construction,” IEEE Trans.
Howard Fan, "Direct Blind Multiuser Detection for CDMA in Multipath without Channel Estimation," IEEE Trans. Milstein, “Performance analysis of MMSE receivers for DS-CDMA in frequency-selective channels,” IEEE Trans. Shensa, “Performance of adaptive linear interference suppression in the presence of dynamic fading,” IEEE Trons.
Li, "Linear Prediction Approach for Joint Blind Equalization and Blind Multiuser Detection in COMA Systems," IEEE Trans. Milstein, "Performance Analysis of MMSE Receivers for OS-COMA in Frequency Selective Channels", IEEE Trans. Clark, "Theoretical Reliability of MMSE Linear Diversity Combining in Raleigh-Fading Additive Interference Channels," IEEE Trans.
